Table 2. Baseline hyperparameters for the CNN+BiLSTM model.

Parameter	Value
Vocabulary size	1,000
Input vector size	53
Dimension of embedding	128
BiLSTM unit size	[65, 45, 40, 35, 20, 15, 10]
Number of convolutional layers	1
Number of hidden layers	3
Number of filters	[6, 9, 10, 16]
Filter size	[7, 8, 10]
Dropout	0.9
Activation function	Softmax
Number of epochs	7
Batch size	[8, 16]